MSCI has officially adopted new index exclusion rules, with Strategy and Metaplanet facing potential exclusion.

According to Lookonchain's monitoring, the whale Hamr4x has once again offloaded 145,010 SOL tokens worth $15.25 million after a two-month hiatus.

South Korea's Ministry of Finance responded to the recent sharp stock market plunge, stating that relevant authorities will take strict measures against market-disrupting activities. Earlier reports said South Korea's financial regulators convened an emergency meeting on the stock market plunge yesterday afternoon, after which the market's decline narrowed.

According to Lookonchain monitoring, a wallet linked to crypto mining firm Hut 8 has deposited 1,589 BTC (valued at $130 million) to a centralized exchange (CEX) over the past two days, likely for sale. The company holds a total of 10,278 BTC, worth $845.6 million.

Beating AI News Flash: Google has begun rolling out Gemini 4 Argon to paid Google AI Pro subscribers. Previously, while Argon was already available in the Gemini app, it was restricted to the pricier Ultra subscription tier within the AI coding tool Antigravity. Now, Pro users can also access this flagship model via Antigravity’s desktop app and CLI to handle complex programming, debugging, and multi-step development tasks. Google has also set a separate weekly usage cap for Argon for Pro users, which does not take up their existing Gemini 3.8 Flash quota, allowing the two models to be used independently. That said, Argon still does not support Antigravity’s built-in IDE, and can only be accessed via the desktop app, CLI, or third-party IDE extensions. Additionally, this rollout is limited to full paid Pro subscriptions; trial subscriptions are not eligible.

Coinbase Assets announced it will suspend trading of Neon EVM (NEON) and Syndicate (SYND) on Coinbase.com, Coinbase Exchange, and Coinbase Prime at approximately 03:00 Beijing Time on November 17. The NEON order book has entered limit-only mode, with users still able to submit, cancel, and match limit orders. NEON and SYND assets remain accessible and withdrawable.

Beating AI Express: Meta has open-sourced the weights of the Muse Spark Open model and partnered with Apple to enable this 816-billion-parameter large model to run natively on Mac Studio. The model activates only 42 billion parameters per inference, supports a 1-million-token context window, and is primarily targeted at programming and complex agent tasks. The weights were made available on Hugging Face on October 15; developers can apply for access after logging in and agreeing to the access terms. Meta and Apple have clashed multiple times in the past over privacy policies, but Meta CEO Mark Zuckerberg has now publicly thanked Apple executive John Ternus, who responded personally. The two companies leveraged Apple’s MLX framework to distribute the model across multiple Mac Studio units for execution. Even when compressed to 4-bit precision, the model requires around 493GB of memory; combined with inference overhead, a single Mac Studio cannot accommodate it. For practical deployment, at least two Mac Studio units equipped with M5 Ultra chips and 512GB of unified memory each are needed, with hardware costs exceeding $40,000. Apple’s tests show that four M5 Ultra Mac Studio units working together can generate roughly 37 tokens per second with an 8192-token context window. Meta has adopted a custom Muse Model License that allows developers to modify, commercialize, and distribute the model, but requires attribution of Muse Spark Open in products or related documentation. For enterprises and their affiliates with total annual revenue exceeding $20 million over 12 consecutive months, a separate agreement with Meta is required if they intend to offer the model as an API service.